342 ✦ Chapter 8: The AUTOREG Procedure GARCH Models The AUTOREG procedure supports several variations of GARCH models. Using the TYPE= option along with the GARCH= option enables you to control the constraints placed on the estimated GARCH parameters. You can specify unconstrained, nonnegativity-constrained (default), stationarity-constrained, or integration-constrained models. The integration constraint produces the integrated GARCH (IGARCH) model. You can also use the TYPE= option to specify the exponential form of the GARCH model, called the EGARCH model, or other types of GARCH models, namely the quadratic GARCH (QGARCH), threshold GARCH (TGARCH), and power GARCH (PGARCH) models. The MEAN= option along with the GARCH= option specifies the GARCH-in-mean (GARCH-M) model. The following statements illustrate the use of the TYPE= option to fit an AR(2)-EGARCH .1; 1/ model to the series Y. (Output is not shown.) / * AR(2)-EGARCH(1,1) model * / proc autoreg data=a; model y = time / nlag=2 garch=(p=1,q=1,type=exp); run; See the section “GARCH Models” on page 375 for details. Syntax: AUTOREG Procedure The AUTOREG procedure is controlled by the following statements: PROC AUTOREG options ; BY variables ; CLASS variables ; MODEL dependent = regressors / options ; HETERO variables / options ; NLOPTIONS options ; RESTRICT equation , . . . , equation ; TEST equation , . . . , equation / option ; OUTPUT OUT = SAS data set options ; At least one MODEL statement must be specified. One OUTPUT statement can follow each MODEL statement. One HETERO statement can follow each MODEL statement. Functional Summary The statements and options used with the AUTOREG procedure are summarized in the following table. Functional Summary ✦ 343 Table 8.1 AUTOREG Functional Summary Description Statement Option Data Set Options Specify the input data set AUTOREG DATA= Write parameter estimates to an output data set AUTOREG OUTEST= Include covariances in the OUTEST= data set AUTOREG COVOUT Requests that the procedure produce graphics via the Output Delivery System AUTOREG PLOTS= Write predictions, residuals, and confidence limits to an output data set OUTPUT OUT= Declaring the Role of Variables Specify BY-group processing BY Specify classification variables CLASS Printing Control Options Request all printing options MODEL ALL Print transformed coefficients MODEL COEF Print correlation matrix of the estimates MODEL CORRB Print covariance matrix of the estimates MODEL COVB Print DW statistics up to order j MODEL DW=j Print marginal probability of the generalized Durbin-Watson test statistics for large sample sizes MODEL DWPROB Print the p-values for the Durbin-Watson test be computed using a linearized approximation of the design matrix MODEL LDW Print inverse of Toeplitz matrix MODEL GINV Print the Godfrey LM serial correlation test MODEL GODFREY= Print details at each iteration step MODEL ITPRINT Print the Durbin t statistic MODEL LAGDEP Print the Durbin h statistic MODEL LAGDEP= Print the log-likelihood value of the regression model MODEL LOGLIKL Print the Jarque-Bera normality test MODEL NORMAL Print the tests for the absence of ARCH effects MODEL ARCHTEST= Print BDS tests for independence MODEL BDS= Print rank version of von Neumann ratio test for independence MODEL VNRRANK= Print runs test for independence MODEL RUNS= Print the turning point test for independence MODEL TP= Print the Lagrange multiplier test HETERO TEST=LM Print the Chow test MODEL CHOW= Print the predictive Chow test MODEL PCHOW= Suppress printed output MODEL NOPRINT Print partial autocorrelations MODEL PARTIAL Print Ramsey’s RESET test MODEL RESET 344 ✦ Chapter 8: The AUTOREG Procedure Table 8.1 continued Description Statement Option Print Phillips-Perron tests for stationarity or unit roots MODEL STATIONARITY=(PHILLIPS=) Print Augmented Dickey-Fuller tests for stationarity or unit roots MODEL STATIONARITY=(ADF=) Print ERS tests for stationarity or unit roots MODEL STATIONARITY=(ERS=) Print Ng-Perron tests for stationarity or unit roots MODEL STATIONARITY=(NP=) Print KPSS tests for stationarity or unit roots MODEL STATIONARITY=(KPSS=) Print tests of linear hypotheses TEST Specify the test statistics to use TEST TYPE= Print the uncentered regression R 2 MODEL URSQ Options to Control the Optimization Process Specify the optimization options NLOPTIONS see Chapter 6, “Nonlinear Optimization Methods,” Model Estimation Options Specify the order of autoregressive process MODEL NLAG= Center the dependent variable MODEL CENTER Suppress the intercept parameter MODEL NOINT Remove nonsignificant AR parameters MODEL BACKSTEP Specify significance level for BACKSTEP MODEL SLSTAY= Specify the convergence criterion MODEL CONVERGE= Specify the type of covariance matrix MODEL COVEST= Set the initial values of parameters used by the iterative optimization algorithm MODEL INITIAL= Specify iterative Yule-Walker method MODEL ITER Specify maximum number of iterations MODEL MAXITER= Specify the estimation method MODEL METHOD= Use only first sequence of nonmissing data MODEL NOMISS Specify the optimization technique MODEL OPTMETHOD= Imposes restrictions on the regression estimates RESTRICT Estimate and test heteroscedasticity models HETERO GARCH Related Options Specify order of GARCH process MODEL GARCH=(Q=,P=) Specify type of GARCH model MODEL GARCH=(: : :,TYPE=) Specify various forms of the GARCH-M model MODEL GARCH=(: : :,MEAN=) Suppress GARCH intercept parameter MODEL GARCH=(: : :,NOINT) Specify the trust region method MODEL GARCH=(: : :,TR) Estimate the GARCH model for the conditional t distribution MODEL GARCH=(: : :) DIST= Functional Summary ✦ 345 Table 8.1 continued Description Statement Option Estimate the start-up values for the conditional variance equation MODEL GARCH=( : : : ,STARTUP=) Specify the functional form of the heteroscedasticity model HETERO LINK= Specify that the heteroscedasticity model does not include the unit term HETERO NOCONST Impose constraints on the estimated parameters in the heteroscedasticity model HETERO COEF= Impose constraints on the estimated standard deviation of the heteroscedasticity model HETERO STD= Output conditional error variance OUTPUT CEV= Output conditional prediction error variance OUTPUT CPEV= Specify the flexible conditional variance form of the GARCH model HETERO Output Control Options Specify confidence limit size OUTPUT ALPHACLI= Specify confidence limit size for structural predicted values OUTPUT ALPHACLM= Specify the significance level for the upper and lower bounds of the CUSUM and CUSUMSQ statistics OUTPUT ALPHACSM= Specify the name of a variable to contain the values of the Theil’s BLUS residuals OUTPUT BLUS= Output the value of the error variance 2 t OUTPUT CEV= Output transformed intercept variable OUTPUT CONSTANT= Specify the name of a variable to contain the CUSUM statistics OUTPUT CUSUM= Specify the name of a variable to contain the CUSUMSQ statistics OUTPUT CUSUMSQ= Specify the name of a variable to contain the upper confidence bound for the CUSUM statistic OUTPUT CUSUMUB= Specify the name of a variable to contain the lower confidence bound for the CUSUM statistic OUTPUT CUSUMLB= Specify the name of a variable to contain the upper confidence bound for the CUSUMSQ statistic OUTPUT CUSUMSQUB= Specify the name of a variable to contain the lower confidence bound for the CUSUMSQ statistic OUTPUT CUSUMSQLB= Output lower confidence limit OUTPUT LCL= Output lower confidence limit for structural predicted values OUTPUT LCLM= Output predicted values OUTPUT P= 346 ✦ Chapter 8: The AUTOREG Procedure Table 8.1 continued Description Statement Option Output predicted values of structural part OUTPUT PM= Output residuals OUTPUT R= Output residuals from structural predictions OUTPUT RM= Specify the name of a variable to contain the part of the predictive error variance (v t ) OUTPUT RECPEV= Specify the name of a variable to contain recursive residuals OUTPUT RECRES= Output transformed variables OUTPUT TRANSFORM= Output upper confidence limit OUTPUT UCL= Output upper confidence limit for structural predicted values OUTPUT UCLM= PROC AUTOREG Statement PROC AUTOREG options ; The following options can be used in the PROC AUTOREG statement: DATA=SAS-data-set specifies the input SAS data set. If the DATA= option is not specified, PROC AUTOREG uses the most recently created SAS data set. OUTEST=SAS-data-set writes the parameter estimates to an output data set. See the section “OUTEST= Data Set” on page 410 later in this chapter for information on the contents of these data set. COVOUT writes the covariance matrix for the parameter estimates to the OUTEST= data set. This option is valid only if the OUTEST= option is specified. PLOTS<(global-plot-options)> < = (specific plot options)> requests that the AUTOREG procedure produce statistical graphics via the Output Delivery System, provided that the ODS GRAPHICS statement has been specified. For general infor- mation about ODS Graphics, see Chapter 21, “Statistical Graphics Using ODS” (SAS/STAT User’s Guide). The global-plot-options apply to all relevant plots generated by the AUTOREG procedure. The global-plot-options supported by the AUTOREG procedure follow. Global Plot Options ONLY suppresses the default plots. Only the plots specifically requested are produced. UNPACKPANEL breaks a graphic that is otherwise paneled into individual component plots. BY Statement ✦ 347 Specific Plot Options ALL requests that all plots appropriate for the particular analysis be produced. ACF produces the autocorrelation function plot. IACF produces the inverse autocorrelation function plot of residuals. PACF produces the partial autocorrelation function plot of residuals. FITPLOT plots the predicted and actual values. COOKSD produces the Cook’s D plot. QQ Q-Q plot of residuals. RESIDUAL | RES plots the residuals. STUDENTRESIDUAL plots the studentized residuals. For the models with the NLAG= or GARCH= options in the MODEL statement or with the HETERO statement, this option is replaced by the STANDARDRESIDUAL option. STANDARDRESIDUAL plots the standardized residuals. WHITENOISE plots the white noise probabilities. RESIDUALHISTOGRAM | RESIDHISTOGRAM plots the histogram of residuals. NONE suppresses all plots. In addition, any of the following MODEL statement options can be specified in the PROC AU- TOREG statement, which is equivalent to specifying the option for every MODEL statement: ALL, ARCHTEST, BACKSTEP, CENTER, COEF, CONVERGE=, CORRB, COVB, DW=, DWPROB, GINV, ITER, ITPRINT, MAXITER=, METHOD=, NOINT, NOMISS, NOPRINT, and PARTIAL. BY Statement BY variables ; A BY statement can be used with PROC AUTOREG to obtain separate analyses on observations in groups defined by the BY variables. CLASS Statement (Experimental) CLASS variables ; The CLASS statement names the classification variables to be used in the analysis. Classification variables can be either character or numeric. In PROC AUTOREG, the CLASS statement enables you to output class variables to a data set that contains a copy of the original data. 348 ✦ Chapter 8: The AUTOREG Procedure Class levels are determined from the formatted values of the CLASS variables. Thus, you can use formats to group values into levels. See the discussion of the FORMAT procedure in SAS Language Reference: Dictionary for details. MODEL Statement MODEL dependent = regressors / options ; The MODEL statement specifies the dependent variable and independent regressor variables for the regression model. If no independent variables are specified in the MODEL statement, only the mean is fitted. (This is a way to obtain autocorrelations of a series.) Models can be given labels of up to eight characters. Model labels are used in the printed output to identify the results for different models. The model label is specified as follows: label : MODEL . . . ; The following options can be used in the MODEL statement after a slash (/). CENTER centers the dependent variable by subtracting its mean and suppresses the intercept parameter from the model. This option is valid only when the model does not have regressors (explanatory variables). NOINT suppresses the intercept parameter. Autoregressive Error Options NLAG=number NLAG=(number-list) specifies the order of the autoregressive error process or the subset of autoregressive error lags to be fitted. Note that NLAG=3 is the same as NLAG=(1 2 3). If the NLAG= option is not specified, PROC AUTOREG does not fit an autoregressive model. GARCH Estimation Options DIST=value specifies the distribution assumed for the error term in GARCH-type estimation. If no GARCH= option is specified, the option is ignored. If EGARCH is specified, the distribution is always the normal distribution. The values of the DIST= option are as follows: T specifies Student’s t distribution. NORMAL specifies the standard normal distribution. The default is DIST=NORMAL. MODEL Statement ✦ 349 GARCH=(option-list) specifies a GARCH-type conditional heteroscedasticity model. The GARCH= option in the MODEL statement specifies the family of ARCH models to be estimated. The GARCH .1; 1/ regression model is specified in the following statement: model y = x1 x2 / garch=(q=1,p=1); When you want to estimate the subset of ARCH terms, such as ARCH .1; 3/ , you can write the SAS statement as follows: model y = x1 x2 / garch=(q=(1 3)); With the TYPE= option, you can specify various GARCH models. The IGARCH .2; 1/ model without trend in variance is estimated as follows: model y = / garch=(q=2,p=1,type=integ,noint); The following options can be used in the GARCH=( ) option. The options are listed within parentheses and separated by commas. Q=number Q=(number-list) specifies the order of the process or the subset of ARCH terms to be fitted. P=number P=(number-list) specifies the order of the process or the subset of GARCH terms to be fitted. If only the P= option is specified, P= option is ignored and Q=1 is assumed. TYPE=value specifies the type of GARCH model. The values of the TYPE= option are as follows: EXP | EGARCH specifies the exponential GARCH or EGARCH model. INTEGRATED | IGARCH specifies the integrated GARCH or IGARCH model. NELSON | NELSONCAO specifies the Nelson-Cao inequality constraints. NONNEG specifies the GARCH model with nonnegativity constraints. POWER | PGARCH specifies the power GARCH or PGARCH model. QUADR | QUADRATIC | QGARCH specifies the quadratic GARCH or QGARCH model. STATIONARY constrains the sum of GARCH coefficients to be less than 1. THRES | THRESHOLD | TGARCH specifies the threshold GARCH or TGARCH model. The default is TYPE=NELSON. 350 ✦ Chapter 8: The AUTOREG Procedure MEAN=value specifies the functional form of the GARCH-M model. The values of the MEAN= option are as follows: LINEAR specifies the linear function: y t D x 0 t ˇ Cıh t C t LOG specifies the log function: y t D x 0 t ˇ Cı ln.h t / C t SQRT specifies the square root function: y t D x 0 t ˇ Cı p h t C t NOINT suppresses the intercept parameter in the conditional variance model. This option is valid only with the TYPE=INTEG option. STARTUP=MSE | ESTIMATE requests that the positive constant c for the start-up values of the GARCH conditional error variance process be estimated. By default or if STARTUP=MSE is specified, the value of the mean squared error is used as the default constant. TR uses the trust region method for GARCH estimation. This algorithm is numerically stable, though computation is expensive. The double quasi-Newton method is the default. MODEL Statement ✦ 351 Printing Options ALL requests all printing options. ARCHTEST ARCHTEST=(option-list) specifies tests for the absence of ARCH effects. The following options can be used in the ARCHTEST=( ) option. The options are listed within parentheses and separated by commas. QLM | QLMARCH requests the Q and Engle’s LM tests. LK | LKARCH requests Lee and King’s ARCH tests. WL | WLARCH requests Wong and Li’s ARCH tests. ALL requests all ARCH tests, namely Q and Engle’s LM tests, Lee and King’s tests, and Wong and Li’s tests. If ARCHTEST is defined without additional suboptions, it requests the Q and Engle’s LM tests. That is,the statement model return = x1 x2 / archtest; is equivalent to the statement model return = x1 x2 / archtest=(qlm); The following statement requests Lee and King’s tests and Wong and Li’s tests: model return = / archtest=(lk,wl); BDS BDS=(option-list) specifies Brock-Dechert-Scheinkman (BDS) tests for independence. The following options can be used in the BDS=( ) option. The options are listed within parentheses and separated by commas. M=number specifies the maximum number of the embedding dimension. The BDS tests with embedding dimension from 2 to M are calculated. M must be an integer between 2 and 20. The default value of the M= suboption is 20. . infor- mation about ODS Graphics, see Chapter 21, “Statistical Graphics Using ODS” (SAS/STAT User’s Guide) . The global-plot-options apply to all relevant plots generated by the AUTOREG procedure distribution. NORMAL specifies the standard normal distribution. The default is DIST=NORMAL. MODEL Statement ✦ 3 49 GARCH=(option-list) specifies a GARCH-type conditional heteroscedasticity model. The GARCH= option